COMPARATIVE VALIDATION ANALYSIS OF BIOLOGY MULTIPLE-CHOICE TESTS OF THE NATIONAL EXAMINATION COUNCILS IN SOUTHWESTERN NIGERIA FROM 2015 TO 2017

Abstract

The study compared the validity and reliability of Biology multiple-choice tests conducted by the National Examination Councils (NECO) across six states of the Southwestern Nigeria for three consecutive years (2015 to 2017). It also compared parameters a, b, and c of the NECO Biology Multiple-choice examinations across the three years. These were with a view to ascertaining the stability of the item parameters of Biology Multiple-choice items used by the National Examination Council to assess the students. The study adopted expost facto research design, since the secondary data was used for the study. The population for the study comprised 363,807students who sat for the NECO Biology paper III from 2015 to 2017 in Southwestern Nigeria. A proportionate sample of 10% was taken across the states from 2015 to 2017. This consisted of 36,383 {(Male=17641 (48.5%) and Female=18742 (51.5%)} students. The instrument for the study consisted of 60 NECO Biology paper III used in 2015 to 2017 and the responses to those questions. The responses to those questions were coded as “1” for right option and “0” for wrong option. The data collected was analysed using Exploratory Factor Analysis (EFA) for validity, Cronbach’s Alpha and Spearman Brown Split-Half to assess the reliability, while one-way analysis of variance was use for comparing students’ scores. The results of the component matrix revealed factor loading ranges from 0.611 to 0.721, which was above the minimum value of 0.6 recommended, as well as a dominant factor underlying the construct was recorded variance greater than 20% in two years of the three consecutive years reviewed in the study. The results also showed high coefficients of reliability using Split-Half yielded 0.94 (94%), 0.91 (91%), 0.95 (95%) and Cronbach’s Alpha yielded 0.92 (92%), 0.91 (91%) and 0.93 (93%) for the three years respectively. Finally, there was no statistical significant difference in the item difficulty (F(2,177)= 0.979; p>0.05), discrimination (F(2,177)= 0.573; p>0.05) and guessing (F(2,177)= 0.984; p>0.05) parameters across the three years. The study concluded that the NECO Biology Paper III was consistently stable in terms of validity, reliability and parameter estimates.